Introduction of a SiFA Moiety into the D-Glutamate Chain of DOTA-PP-F11N Results in Radiohybrid-Based CCK-2R-Targeted Compounds with Improved Pharmacokinetics In Vivo

In order to enable 18F- and 177Lu-labelling within the same molecule, we introduced a silicon-based fluoride acceptor (SiFA) into the hexa-D-glutamate chain of DOTA-PP-F11N. In addition, minigastrin analogues with a prolonged as well as γ-linked D-glutamate chain were synthesised and evaluated. CCK-2R affinity (IC50, AR42J cells) and lipophilicity (logD7.4) were determined. Biodistribution studies at 24 h post-injection (p.i.) and µSPECT/CT imaging at 1, 4 and 24 h p.i. were carried out in AR42J tumour-bearing CB17-SCID mice. CCK-2R affinity of (R)-DOTAGA-rhCCK-1 to 18 was enhanced with increasing distance between the SiFA building block and the binding motif. Lipophilicity of [177Lu]Lu-(R)-DOTAGA-rhCCK-1 to 18 was higher compared to that of [177Lu]Lu-DOTA-PP-F11N and [177Lu]Lu-CP04. The respective α- and γ-linked rhCCK derivatives revealing the highest CCK-2R affinity were further evaluated in vivo. In comparison with [177Lu]Lu-DOTA-PP-F11N, [177Lu-]Lu-(R)-DOTAGA-rhCCK-9 and -16 exhibited three- to eight-fold increased activity levels in the tumour at 24 h p.i. However, activity levels in the kidneys were elevated as well. We could show that the introduction of a lipophilic SiFA moiety into the hydrophilic backbone of [177Lu]Lu-DOTA-PP-F11N led to a decelerated blood clearance and thus improved tumour retention. However, elevated kidney retention has to be addressed in future studies.


Introduction
Medullary thyroid carcinoma (MTC) constitutes for only 2-3% of all thyroid cancer cases and is therefore rather rare, but treatment options are limited: neither external beam radiation, nor conventional chemotherapy, nor radioiodine therapy are recommended, as all three concepts have not shown curative effects [1][2][3][4]. Tyrosine kinase inhibitors such as selpercatinib, vandetanib or cabozantinib are usually applied for systematic treatment but these agents are associated with distinct side effects such as renal toxicity, myelosuppression, arterial thromboembolism, hepatotoxicity, and muscle wasting [3,5].
Since Reubi et al. discovered that approximately 92% of all MTCs overexpress the cholecystokinin-2 receptor (CCK-2R), designing small compounds that address this target became attractive in combination with peptide receptor radionuclide imaging (PRRI) and therapy (PRRT) [6]. While first compounds were based on the structure of cholecystokinin, nowadays minigastrin-based ligands are clearly favoured because of their increased hydrophilicity [7]. However, early radiolabelled minigastrin analogues suffered from elevated activity levels in the kidneys, which hampered a potential therapeutic use [8,9].
An important step for the applicability of these minigastrin derivatives was the modification within the linker section, namely the substitution of the hexa-L-glutamate by a kinin, nowadays minigastrin-based ligands are clearly favoured because of their increased hydrophilicity [7]. However, early radiolabelled minigastrin analogues suffered from elevated activity levels in the kidneys, which hampered a potential therapeutic use [8,9]. An important step for the applicability of these minigastrin derivatives was the modification within the linker section, namely the substitution of the hexa-L-glutamate by a hexa-D-glutamate chain, which resulted in compounds such as CP04 and DOTA-PP-F11N, amongst others. The latter consists of a stabilised binding motif of seven amino acid with high CCK-2R affinity (H-Ala-Tyr-Gly-Trp-Nle-Asp-Phe-NH2), a hexa-D-glutamate linker and DOTA (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid) as a chelator [10,11]. Nevertheless, due to the high hydrophilicity of [ 177 Lu]Lu-DOTA-PP-F11N, first patient studies revealed a very rapid renal clearance already at 1 h post-injection (p.i.), which resulted in a median (interquartile range) absorbed tumour dose of only 0.88 Gy/GBq [12]. Moreover, none of the currently available CCK-2R-targeted compounds for clinical application bears an option for 18 F-labelling. Recently, radiohybrid (rh)-based prostate-specific membrane antigen (PSMA)-targeted compounds were developed by our group, implementing a new class of theranostic compounds. These compounds comprise a silicon-based fluoride acceptor (SiFA) moiety for rapid and facile 18 F-fluorination via a 18 F/ 19 F isotopic exchange reaction and additionally contains a chelator for radiometallation (with 68 Ga or 177 Lu, amongst others). This concept results in a chemically identical pair of compounds (either 18 F/non-radioactive metal or 19 F/radiometal), which thus exhibits identical pharmacokinetics and can be used for either diagnostic or therapeutic applications [13,14].

Synthesis and Radiolabelling
The uncomplexed ligands were synthesised via standard Fmoc-based SPPS, yielding 5-20% RP-HPLC purified precursors (chemical purity > 95%, determined by RP-HPLC at Pharmaceuticals 2022, 15, 1467 3 of 12 λ = 220 nm). Non-radioactive labelling proceeded quantitatively using a 2.5-fold excess of [ nat Lu]LuCl 3 . No purification prior to affinity studies was performed, as the remaining free Lu 3+ was shown to not affect affinity data [19]. 177 Lu-labelling of all compounds was carried out manually resulting in quantitative radiochemical yields and purities of >95% as well as molar activities of 30 ± 10 GBq/µmol. After radiolabelling all peptides were used without further purification. Confirmation of peptide integrity and quality controls are depicted in the Supplementary Materials ( Figures S1-S3).

In Vitro Characterisation
The affinity and lipophilicity data of all compounds are summarised in Figure 2 and Table S1.

Synthesis and Radiolabelling
The uncomplexed ligands were synthesised via standard Fmoc-based SPPS, yielding 5-20% RP-HPLC purified precursors (chemical purity >95%, determined by RP-HPLC at λ = 220 nm). Non-radioactive labelling proceeded quantitatively using a 2.5-fold excess of [ nat Lu]LuCl3. No purification prior to affinity studies was performed, as the remaining free Lu 3+ was shown to not affect affinity data [19]. 177 Lu-labelling of all compounds was carried out manually resulting in quantitative radiochemical yields and purities of >95% as well as molar activities of 30 ± 10 GBq/µ mol. After radiolabelling all peptides were used without further purification. Confirmation of peptide integrity and quality controls are depicted in the Supplementary Materials ( Figures S1-S3).

In Vitro Characterisation
The affinity and lipophilicity data of all compounds are summarised in Figure 2 and Table S1. In general, all ligands containing a γ-linked D-glutamate chain revealed a higher affinity towards CCK-2R compared to their α-linked counterparts, except for [ nat Lu]Lu-10. Furthermore, a trend could be observed that with increasing distance of the dap(SiFA) moiety to the binding motif IC50 values decreased, irrespective whether the compounds In general, all ligands containing a γ-linked D-glutamate chain revealed a higher affinity towards CCK-2R compared to their α-linked counterparts, except for [ nat Lu]Lu-10.  All four reference ligands revealed a high hydrophilicity, exhibiting distribution coefficients (logD 7.4 ) in a range of −4.8 and −3.8. Not surprisingly, the rhCCK derivatives comprising the lipophilic SiFA moiety displayed a distinctly higher lipophilicity (logD 7.4 = −2.9 to −1. 7).

Discussion
Among the currently most promising minigastrin-derived peptides for clinical application (DOTA-MGS5, DOTA-PP-F11N and CP04), there is no option available for facile 18 Flabelling [20][21][22][23]. Particularly a 18 F-labelled minigastrin analogue would be beneficial for the detection of small MTC-derived metastases due to the low tissue penetration and thus high resolution of fluorine-18, consequently to its low positron energy (E β,max = 635 keV), compared to gallium-68, for example [24]. Furthermore, up to now there is no CCK-2Rtargeted ligand that enables both 18 F-and 177 Lu-labelling. This radiohybrid-based concept has been successfully implemented for PSMA inhibitors and resulted in impressive results over the last three years, which is why the compounds rhPSMA-7.3 and rhPSMA-10.1 are evaluated in clinical studies [13,[25][26][27][28][29][30][31][32]. With the aim to develop a 177 Lu-labelled minigastrin analogue, which can also be labelled with fluorine-18, we introduced the lipophilic SiFA moiety into different sites within the highly hydrophilic, N-terminal D-glutamate chain of DOTA-PP-F11N and compared these novel compounds to the reference compounds, The SiFA moiety was introduced into different sites within both an α-linked poly-Dglutamate chain (= (R)-DOTAGA-rhCCK-1 to -9) and a γ-linked poly-D-glutamate chain (= (R)-DOTAGA-rhCCK-10 to -18), while the binding sequence of DOTA-PP-F11N was maintained. In general, for both linker concepts it could be observed that with increasing distance of the SiFA moiety to the binding motif, CCK-2R affinity was enhanced, which indicates that the bulky SiFA building block does not fit into the binding pocket of the receptor. However, at a farther distance the SiFA group seems to be located outside of the binding pocket and thus CCK-2R affinity increases. In direct comparison of these series of minigastrin derivatives (α-or γ-linked D-glutamate chain) that each exhibit the same site for the SiFA moiety, it is evident that those ligands containing a γ-linked D-glutamate chain generally show a higher CCK-2R affinity. Similar results were observed for the respective γ-linked analogues of the references, [ nat Lu]Lu-DOTA-PP-γ-F11N and [ nat Lu]Lu-γ-CP04, pointing to a beneficial effect of a prolonged linker section and thus the use of γ-linked D-glutamate residues. Nevertheless, IC 50 values of the most affine compounds were still approximately fivefold (α-linked poly-D-glutamate linker) and twofold (γ-linked poly-Dglutamate linker) higher compared to [ nat Lu]Lu-DOTA-PP-F11N and [ nat Lu]Lu-CP04 (IC 50 of 11-13 nM). Hence, it is assumed that the lower CCK-2R affinity of the novel rhCCK derivatives is due either to the addition of the sterically demanding SiFA moiety or the (R)-DOTAGA chelator, which comprises one negative charge more than the DOTA chelator present in the reference compounds when labelled with [ nat/177 Lu]lutetium.
The addition of the SiFA moiety was also accompanied by an enhanced lipophilicity (logD 7.4 : −2.9 to −1.7), which is one to three magnitudes higher compared to [ 177 Lu]Lu-DOTA-PP-F11N and [ 177 Lu]Lu-CP04. However, this was desired since for tumour targeting we consider a logD 7.4 value of about −4 or lower unfavourable because we assume that high tumour uptake is prevented by a too rapid clearance rate. Indeed, first patient studies with [ 177 Lu]Lu-DOTA-PP-F11N showed low activity levels in the tumour but high levels in the bladder at 1 h p.i., most likely due to an accelerated clearance [12]. Based on previous experiences in our group, a range of −3 to −2 seems to be ideal for tumour targeting.
We thus selected the most promising α-and γ-linked compound with regard to CCK-2R affinity (IC 50 ) and lipophilicity (logD 7.4 ), (R)-DOTAGA-rhCCK-9 and -16 ( Figure 6 [ 177 Lu]Lu-(R)-DOTAGA-rhCCK-9 revealed low internalisation values, which is in accordance with its low CCK-2R affinity but contradicting to the higher tumour uptake found compared to the reference. It has to be added that tumour values for [ 177 Lu]Lu-DOTA-PP-F11N were low despite its high CCK-2R affinity, which is, however, in accordance to the patient data and most likely caused its high hydrophilicity and thus rapid clearance [12]. Although further studies have to be carried out to elucidate the beneficial effects observed in this study, we proved that the introduction of the SiFA group not only generated a possibility for 18 F-labelling but also improved overall bioavailability in vivo. Besides the decreased hydrophilicity, we further suspect an elevated albumin binding potential of the SiFA group, which decelerates the activity clearance, increases circulation time of the compounds in the blood and thus enhances activity uptake and retention in the tumour. Similar observations were made for PSMA-targeted compounds [14,33].
spective γ-linked analogues of the references, [ Lu]Lu-DOTA-PP-γ-F11N and [ Lu]Luγ-CP04, pointing to a beneficial effect of a prolonged linker section and thus the use of γlinked D-glutamate residues. Nevertheless, IC50 values of the most affine compounds were still approximately fivefold (α-linked poly-D-glutamate linker) and twofold (γ-linked poly-D-glutamate linker) higher compared to [ nat Lu]Lu-DOTA-PP-F11N and [ nat Lu]Lu-CP04 (IC50 of 11-13 nM). Hence, it is assumed that the lower CCK-2R affinity of the novel rhCCK derivatives is due either to the addition of the sterically demanding SiFA moiety or the (R)-DOTAGA chelator, which comprises one negative charge more than the DOTA chelator present in the reference compounds when labelled with [ nat/177 Lu]lutetium.
The addition of the SiFA moiety was also accompanied by an enhanced lipophilicity (logD7.4: −2.9 to −1.7), which is one to three magnitudes higher compared to [ 177 Lu]Lu-DOTA-PP-F11N and [ 177 Lu]Lu-CP04. However, this was desired since for tumour targeting we consider a logD7.4 value of about −4 or lower unfavourable because we assume that high tumour uptake is prevented by a too rapid clearance rate. Indeed, first patient studies with [ 177 Lu]Lu-DOTA-PP-F11N showed low activity levels in the tumour but high levels in the bladder at 1 h p.i., most likely due to an accelerated clearance [12]. Based on previous experiences in our group, a range of −3 to −2 seems to be ideal for tumour targeting.
Despite these respectable results, our current rhCCK derivatives suffer from elevated activity levels in the kidneys (30-fold higher compared to the reference). We assume a synergistic effect of the negative charges in proximity of the SiFA moiety within the linker section, as [ 177 Lu]Lu-DOTA-PP-F11N did not show comparable kidney values although it comprises a similar amount of negative charges in its linker. These kidney issues have to be addressed in future studies to enable a clinical translation of this rh-based concept for minigastrin analogues. One possible strategy to decrease kidney accumulation and retention might be a reduction of the albumin binding of the rhCCK derivatives, as beneficial effects were observed for PSMA inhibitors when negative charges in direct proximity to the SiFAbuilding block were depleted [14]. Furthermore, co-injection of lysine or gelofusine could also be a valuable tool. Tumour values were noticeably higher at 1 and 4 h p.i. compared to the reference in µSPECT/CT imaging (n = 1). However, further studies at 1 and 4 h p.i. have to be conducted to statistically confirm these observations and, furthermore, elucidate the imaging potential of the respective 18 F-labeled rhCCK analogues.
In summary, we could successfully introduce a SiFA building block into the minigastrin analogue DOTA-PP-F11N, which not only generated a possibility for 18 F-labelling but also considerably improved pharmacokinetics. We further could show that the rh-based concept successfully applied for PSMA-targeted compounds can be applied for CCK-2Rtargeted ligands as well, which enables both 18 F-and 177 Lu-labelling for a theranostic use. Nevertheless, elevated activity levels in the kidneys are of concern, which has to be optimised in future studies. Moreover, CCK-2R affinity might possibly be further improved, either by varying the position of the SiFA building block or a DOTA-for-(R)-DOTAGA substitution. However, a beneficial effect of a γ-instead of an α-linked D-glutamate chain in minigastrin derivatives was found, which might be applicable for other peptides and their linker as well.

Materials and Methods
Characterisation of all CCK-2R-targeted compounds is provided in the Supplementary Materials ( Figures S1-S4). Electrospray ionisation-mass spectra for characterisation of the substances were acquired on an expression L CMS mass spectrometer (Advion Ltd., Harlow, UK).

Chemical Synthesis and Labelling Procedures
All compounds were synthesised via standard Fmoc-based solid phase peptide synthesis (SPPS) using a H-Rink Amide ChemMatrix ® resin (35-100 mesh particle size, 0.4-0.6 mmol/g loading, Merck KGaA, Darmstadt, Germany). Final purification of the peptides was performed by reversed phase high performance liquid chromatography (RP-HPLC). 177 Lu-and nat Lu-complexation of the peptides was performed according to a previously published procedure [14].

In Vivo Experiments
All animal experiments were conducted in accordance with general animal welfare regulations in Germany (German animal protection act, in the edition of the announcement, dated 18 May 2006, as amended by Article 280 of 19 June 2020, approval no. ROB-55.2-1-2532.Vet_02-18-109 by the General Administration of Upper Bavaria) and the institutional guidelines for the care and use of animals. CB17-SCID mice of both genders and aged 2-12 months (Charles River Laboratories International Inc., Sulzfeld, Germany) were allowed to acclimate at the in-house animal facility for at least one week prior to tumour cell inoculation was performed. Tumour xenografts were generated using AR42J cells (5.0 × 10 6 cells per 200 µL) suspended in a 1/1 mixture (v/v) of RPMI 1640 medium and Cultrex ® Basement Membrane Matrix Type 3 (Trevigen, Gaithersburg, MD, USA). This suspension was inoculated subcutaneously onto the right shoulder and animals were used when tumour volume was >100 mm 3 (1-2 week after inoculation). Exclusion criteria for animals from an experiment were either weight loss higher than 20%, a tumour size above 1500 mm 3 , an ulceration of the tumour, respiratory distress or a change of behaviour. None of these criteria applied to any animal from the experiment. Neither randomisation nor blinding was applied in the allocation of the experiments. Health status is SPF according to FELASA.
Imaging studies were carried out according to a recently published protocol [19]. Static images were recorded at t = 1, 4 and 24 h p.i. (anesthesia by 2% isoflurane, n = 1) with an acquisition time of t + (45-60 min) using a high-energy general-purpose rat and mouse collimator via MILabs acquisition software v11.00 and v12.26 from MILabs (Utrecht, The Netherlands).
Acquired data were statistically analysed by performing a Student's t-test via Excel (Microsoft Corporation, Redmond, WA, USA) and OriginPro software (version 9.7) from OriginLab Corporation (Northampton, MA, USA). Acquired p values of <0.05 were considered statistically significant.

Conclusions
We could demonstrate that the radiohybrid-based concept could easily be transferred to minigastrin derivatives, whose hydrophilic linker section compensates for the high lipophilicity of the introduced SiFA moiety. This offers not only the possibility of 18 Fand 177 Lu-labelling with the same molecule but also had a beneficial impact on overall pharmacokinetics, as clearance kinetics were decelerated. Thereby, activity retention in the tumour could be increased by approximately eightfold compared to the clinically applied [ 177 Lu]Lu-DOTA-PP-F11N. However, these compounds also suffer from a noticeably enhanced kidney retention. This will be addressed in further studies.